Speech monologues recorded in naturalistic settings provide opportunities to characterize mental illness phenomenology and detect symptom exacerbation. Large language models (LLMs) offer new possibilities for automating this process, as they require annotated data primarily for evaluation rather than training. In this paper, we present a novel automated, multi-agent LLM pipeline for the fine-grained, multi-label extraction of language suggestive of delusional beliefs, associated affective responses, and behavioral responses from transcripts of naturalistic audio diaries collected from people with moderate persecutory ideation. Evaluating an ensemble of three foundation models, we demonstrate that detailed diagnostic prompt instructions successfully reduce false positives for delusional theme classification, but also constrain the interpretation of affective or behavioral responses. Furthermore, comparing multi-agent adjudication frameworks reveals a critical divergence from standard NLP benchmarks: complex conversational debate between agents diminishes accuracy on clinically ambiguous text by inducing premature consensus. Instead, majority voting establishes robust performance (Micro F1 of 0.872 and 0.779 for delusion detection and classification respectively). This work provides a validated and scalable pipeline for the automated detection and characterization of content suggesting delusional beliefs in naturalistic speech.

Cognitive distortions are counterproductive patterns of thinking that are one of the targets of cognitive behavioral therapy (CBT). These can be challenging for clinicians to detect, especially those without extensive CBT training or supervision. Text classification methods can approximate expert clinician judgment in the detection of frequently occurring cognitive distortions in text-based therapy messages. However, performance with infrequent distortions is relatively poor. In this study, we address this sparsity problem with two approaches: Data Augmentation and Domain-Specific Model. The first approach includes Easy Data Augmentation, back translation, and mixup techniques. The second approach utilizes a domain-specific pretrained language model, MentalBERT. To examine the viability of different data augmentation methods, we utilized a real-world dataset of texts between therapists and clients diagnosed with serious mental illness that was annotated for distorted thinking. We found that with optimized parameter settings, mixup was helpful for rare classes. Performance improvements with an augmented model, MentalBERT, exceed those obtained with data augmentation.

There is growing evidence that mobile text message exchanges between patients and therapists can augment traditional cognitive behavioral therapy. The automatic characterization of patient thinking patterns in this asynchronous text communication may guide treatment and assist in therapist training. In this work, we automatically identify distorted thinking in text-based patient-therapist exchanges, investigating the role of conversation history (context) in distortion prediction. We identify six unique types of cognitive distortions and utilize BERT-based architectures to represent text messages within the context of the conversation. We propose two approaches for leveraging dynamic conversation context in model training. By representing the text messages within the context of the broader patient-therapist conversation, the models better emulate the therapist’s task of recognizing distorted thoughts. This multi-turn classification approach also leverages the clustering of distorted thinking in the conversation timeline. We demonstrate that including conversation context, including the proposed dynamic context methods, improves distortion prediction performance. The proposed architectures and conversation encoding approaches achieve performance comparable to inter-rater agreement. The presence of any distorted thinking is identified with relatively high performance at 0.73 F1, significantly outperforming the best context-agnostic models (0.68 F1).